convert.annot.R

require(data.table)
require(stringr)

# GTF TO BED
gtf_to_bed <- function(gtf, feature="transcript", bed_file=NULL) {
  if ("character" %in% class(gtf)) {
    gtf <- fread(gtf)
  } else if ("data.frame" %in% class(gtf)) {
    setDT(gtf)
  }
  setnames(gtf,paste0("V",1:ncol(gtf)))
  bed <- gtf[V3==feature][
    ,V10:=str_extract(V9,sprintf('(?<=%s_id ")[^"]+',feature))][
      ,.(V1,V4,V5,V10,V6,V7)][]
  
  if (!is.null(bed_file)) (
    fwrite(bed, bed_file, sep="\t", col.names = FALSE, quote = FALSE)
  )
  return(bed)
}

# GTF TO GFF
gtf_to_gff <- function(gtf, feature=c("transcript","exon"), gff_file=NULL) {
  if ("character" %in% class(gtf)) {
    gtf <- fread(gtf)
  } else if ("data.frame" %in% class(gtf)) {
    setDT(gtf)
  }
  setnames(gtf,paste0("V",1:ncol(gtf)))
  gff <- gtf[V3 %in% feature]
  for (i in seq_along(feature)) {
    ft = feature[i]
    if (any(grepl(ft, gff[V3==ft,V9])))
      gff[V3==ft, V10:=sprintf("ID=%s",str_extract(V9,sprintf('(?<=%s_id ")[^"]+',ft)))]
    if (i>1) {
      fp = feature[i-1]
      gff[V3==ft, V10:=sprintf("Parent=%s",str_extract(V9,sprintf('(?<=%s_id ")[^"]+',fp)))]
    }
  }
  gff[,V9:=NULL]
  if (!is.null(gff_file)) (
    fwrite(gff, gff_file, sep="\t", col.names = FALSE, quote = FALSE)
  )
  return(gff)
  
}